The Bayesian Advisor Project: Modeling Academic Advising

An academic advisor's job requires that the advisor knows university and departmental requirements; available courses to satisfy those requirements, and a method for predicting relative student success for various course options. Such predictions are necessarily uncertain, both because the advisor has limited information about the student and because student performance can be in-uenced by unforeseen factors. In this work we present a preliminary Bayes Net model of academic advising. In it, the advisor's predictions are based on student performance in courses so far, as described in a student transcript. We describe the underlying model and some initial experiments.